It appears that many more people use Fluent on Linux boxes versus Windows boxes. Why is this? Is it because Fluent runs better on Linux (or Unix) and is somehow better than Windows? Is it because the majority of the fluent user base is more familiar with Linux\Unix than Windows? What about memory (Ram) considerations? Can Linux make more memory available for fluent computations than a Windows machine? What about networking? Is parallel processing easier to setup on Linux\Unix boxes than Windows? Someone please explain. Thanx in advance... Derick

We use both linux and windows for FLUENT. For many models, both perform just fine. Some observations:

1. Linux is more stable for large models, mostly do the better memory management compared to windows. Also, windows won't MALLOC more than 2GB to a process regardless of the RAM in the machine and will choke on a really big model.

2. If you plan to run on any sort of a cluster, linux is the way to go. Clustering in windows has come a long way, but still has a way to go to be competitive.

3. I think FLUENT does all of their development on linux. We have had instances in the past where bugs in the windows version of Gambit didn't exist in the linux version. Most likely, the bugs were introduced during the port to windows.

4. There is no 64-bit version of FLUENT available for windows at this point. We have found throughput improvements of 0-20% from the 64-bit version of FLUENT on linux compared to the 32-bit version on windows. Gambit is 32-bit, so it isn't affected.

[sozen]

One can read some scientific research on the topic here:
https://sciarena.com/storage/models/...-power-of-.pdf

Spoiler:
Linux is better!